A Degradation Feature Extraction Method for Hydraulic Pumps Based Upon MUWDF and MF-DFA
- PDF / 1,277,511 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 10 Downloads / 160 Views
TECHNICAL ARTICLE—PEER-REVIEWED
A Degradation Feature Extraction Method for Hydraulic Pumps Based Upon MUWDF and MF-DFA Jian Sun . Hongru Li . Baohua Xu
Submitted: 19 January 2016 / in revised form: 30 March 2016 / Published online: 7 June 2016 ASM International 2016
Abstract Hydraulic pump degradation feature extraction is a key step of condition-based maintenance. Since vibration signals of hydraulic pumps during degradation are strongly nonlinear and the feature information is too weak to be effectively extracted, a method based upon MUWDF and MF-DFA is proposed. Initially, the MUWDF is presented to reduce disturbances and improve feature information. Approximate signals of various decomposition layers are selected by feature energy factor and fused according to the presented fusion rules. On this basis, the fused signal is further processed by MF-DFA with a sliding window. Multi-fractal spectrum sensitive factors are selected to be the degradation feature vector of the hydraulic pump. The proposed method is verified by vibration signals sampled in a hydraulic pump degradation experiment. Keywords Degradation feature extraction MUWDF MF-DFA Hydraulic pump
Introduction Being one of the important components of a hydraulic system, the hydraulic pump has direct effects on reliability of the whole system [1]. Because of structural characteristics, such as liquid compressibility, the coupling effect between fluid and solid, and inherent mechanical vibrations, the vibration signal of a hydraulic pump during degradation is complicated and nonlinear, such that feature J. Sun (&) H. Li B. Xu Mechanical Engineering College, No. 97, Heping West Road, Shijiazhuang 050003, People’s Republic of China e-mail: [email protected]
information is hard to extract [2]. Consequently, in order to improve prognostic accuracy for condition-based maintenance, an effective method is required for extracting appropriate fault features. The information fusion technique is able to deal with the redundant information and complementary information for improving reliability [3]. Considering the pump structural characteristics, an appropriate fusion algorithm is needed for extracting effective feature information by fusion of multi-channel vibration signals. Currently, various fusion algorithms have been applied in signal analysis and fault diagnosis. However, in the weighted fusion algorithm, the adjustment of the fusion weights is mainly decided by subjective experiences [4, 5]. The Kalman filtering algorithm lacks strict filtering functions for nonlinear systems [6, 7]. The wavelet analysis fusion algorithm may neglect the feature information during the sampling operation [8, 9]. Consequently, the fusion performances based on conventional algorithms are not sufficient and the fault feature is hard to effectively extract, which makes it difficult to meet the requirements of precise diagnosis and prognostics. Being an efficient nonlinear signal processing method, the morphological undecimated wavelet decomposition (MUWD) algorithm omi
Data Loading...
